CALCULATION OF ALERT AND DETECTION OF WILDFIRE: AN ANALYSIS IN THE EASTERN AMAZON

Abstract

Remote sensing and fire risk index are forms of detection and prediction of wildfire commonly used by the National Institute for Space Research to monitoring the frequency of fires in Brazil. However, fire risk index does not detect the wildfire occurrence in some cases, particularly in Eastern Amazon. Thus, this research was motivated by the objective of analyzing the space / time pattern of the fires detected by satellite and the efficiency of the fire prediction alerts generated by fire risk calculation, considering the variability of these meteorological phenomena that can influence the rainfall regime and change the occurrence of wildfires in the region of Eastern Amazonia. The analyses of this research considered the changes in the meteorological conditions modulated by the interannual variability of the phenomenon El Niño / Southern Oscillation occurred between the years of 2000 to 2017. The results showed that in less rainy years the fires were more frequent throughout the period, while, in rainier years, tended to be more concentrated in the second semester, in a higher proportion than that observed in drier years. The largest portion of the wildfires (80%) ococurred in areas altered by human activity”, this result is not perceived by fire alerts, ddue ti the fact that only environmental variables are considered in the calculation, not considering human activity as a parameter, therefore limiting the accuracy in anticipating wildfire occurrencent.